Investigation of Protection Failure Methods in WDM and Elastic Optical Networks

Mohammad Majid Author

08/11/2020 Added

119 Plays

Description

Video Poster Presentation 2020

Searchable Transcript

Search:

[00:00:00.660]Hi, my name is Mohammad Majid and I'll go on
[00:00:03.140]I'll be going over my poster
[00:00:04.300]on the investigation of protection failure methods
[00:00:06.700]in WDM and elastic optical networks.
[00:00:09.430]I conducted this research with professor Byrav Ramamurthy
[00:00:12.290]from the department of Computer Science and Engineering
[00:00:14.700]from the university of Nebraska Lincoln.
[00:00:17.550]So a brief introduction here,
[00:00:18.960]optical networks are communication based networks
[00:00:21.460]that utilize the fiber links and nodes
[00:00:23.770]to traverse services throughout the typology.
[00:00:26.510]WDM optical networks
[00:00:27.880]or wavelength division multiplexing optical networking
[00:00:31.400]allows multiple wavelengths channels
[00:00:33.631]to share a single fiber link.
[00:00:37.060]This increases a transmission of data rates,
[00:00:39.500]but this is this occurs
[00:00:41.120]and performs in a fixed grid scenario.
[00:00:43.410]With the advancements that the WDM provides
[00:00:45.400]the increase of heterogeneous bandwidth demand
[00:00:48.010]and the problem with a fixed grid spectrum assignment
[00:00:52.700]motivates a study of elastic optical networks.
[00:00:57.210]WDM utilizes the routing
[00:00:58.870]and wave in wavelength to solve a problem
[00:01:00.980]while elastic optical networks focuses on the routing
[00:01:04.120]and spectrum of solve a problem.
[00:01:06.300]In elastic optical networks,
[00:01:07.870]each request is given a specific amount of spectrum
[00:01:10.750]in regard to the battery demand of the request
[00:01:13.640]assisted through bandwidth variable transponders.
[00:01:16.810]The purpose of this study
[00:01:18.470]is to investigate the problem of signaling failures.
[00:01:23.550]A signaling failure is unpredictable
[00:01:25.670]and it does occur in a network
[00:01:27.760]then it can redirect the network
[00:01:29.700]while bringing major data loss,
[00:01:31.910]especially for any services
[00:01:33.200]traversing through a failed link.
[00:01:35.470]To tackle such a problem, we investigate different forms
[00:01:38.560]of protection failure methods
[00:01:40.760]while pro proposing a solution ourselves.
[00:01:43.830]The solution is a unicast segment based algorithm
[00:01:46.620]also called the USP algorithm.
[00:01:49.760]Another big motivation is that segment based protection
[00:01:52.830]has an answer for both the lacking parts
[00:01:54.730]of Lincoln path-based protection.
[00:01:57.900]Going over the main constraints on elastic optical networks,
[00:02:01.380]spectrum continued continually constraint
[00:02:03.315]is where each fiber link from the working path
[00:02:06.039]must use the same spectrum throughout.
[00:02:10.640]The spectrum of continually constraints
[00:02:12.094]rise to the adjacency of each spectrum assigned.
[00:02:16.560]The adjacency of spectrum assigned for each fiber link
[00:02:18.970]in the working path.
[00:02:22.360]So each fiber link in a working path
[00:02:25.390]must have adjacent spectrum.
[00:02:27.110]The same spectrum constraint
[00:02:28.320]which is mainly used on Lincoln path based
[00:02:30.930]is where the working and backup paths
[00:02:32.780]will use the same spectrum
[00:02:34.610]under the spectrum contiguity and continuity constraints.
[00:02:38.860]Going over our proposed algorithm
[00:02:41.310]we see that the first step here, of our proposal approach
[00:02:45.704]will be the inputs of a graph
[00:02:48.490]and while we the request will be sources and destination,
[00:02:51.060]along with a bandwidth demand and gigabits per second.
[00:02:54.350]We would like to convert the bandwidth demand
[00:02:56.090]to many gigabits per second,
[00:02:57.500]to F amounts of continuous frequency slots.
[00:03:01.710]We would like to check the blocks for each,
[00:03:04.110]we would like to check the blocks for each fiber link
[00:03:06.030]and check to see if F slots are available
[00:03:08.600]and if not,
[00:03:10.460]then we will increment index by one
[00:03:14.150]and start the process all over again.
[00:03:17.470]And we will always start at the first index.
[00:03:21.240]So we were created G prime graph
[00:03:23.390]where all the edges and a G prime
[00:03:25.120]will come from the edges in the original graph G
[00:03:28.270]that contain the slots F continuous
[00:03:31.910]will contain F amount of continuous slots.
[00:03:35.170]And if no edges are available from the first index again,
[00:03:37.870]we will start the process all over again
[00:03:39.449]and once we do find our G prime graph,
[00:03:41.840]we'll perform the shortest path algorithm
[00:03:43.750]to be dextrous algorithm.
[00:03:45.890]And we will store that working path
[00:03:48.460]for the request in a set.
[00:03:51.980]Of each working path
[00:03:53.870]and for each working path,
[00:03:56.310]we'll check the amount of edges in that working path.
[00:03:59.480]You wanna divide that by two
[00:04:01.730]and if it's divisible by two, then that will be our amounts
[00:04:06.000]we'll divide it up by two
[00:04:07.990]and that will be our amounts of segments
[00:04:11.490]with that working part if it is divisible by two.
[00:04:14.166]Essentially, we would like to store that in a specific set
[00:04:16.670]for that specific request.
[00:04:19.550]You wanna check to see if we also check
[00:04:22.265]that each segment will protect to a specific fiber links
[00:04:26.400]starting from the first index, the working path.
[00:04:29.090]And if it's not, and if the amount of edges
[00:04:31.150]in a working path are not divisible by two
[00:04:33.330]then we will do the same process
[00:04:35.050]except we will take the floor function
[00:04:36.660]in where the last segment
[00:04:37.790]will protect the last three fiber links
[00:04:39.300]in the network and the working path.
[00:04:42.130]There are only two fiber links available
[00:04:44.520]or edges in a working path
[00:04:45.950]there will be strictly two
[00:04:49.440]fiber, excuse me, segments available.
[00:04:52.070]Where the first node will protective to the next node
[00:04:53.940]and next node up to the last node
[00:04:55.850]where all the segments we do provide
[00:04:57.610]will be non overlapping based.
[00:04:59.550]And if only one edge exist in a working path,
[00:05:02.390]the rule is take the shortest path algorithm
[00:05:04.440]find the destroy path and store that in a set
[00:05:07.100]specifically for the backup path of that request.
[00:05:10.890]And assign the necessary slots needed
[00:05:13.480]for working your backup path, using the first method.
[00:05:16.760]And if no paths are found,
[00:05:18.200]they we'll increment the index by one,
[00:05:20.070]to look up the
[00:05:21.910]F amount of continuous frequency slots available.
[00:05:26.090]For each segments of the request
[00:05:28.925]we want to, again, perform shortest path dextrous algorithm
[00:05:31.850]and find this back up this short path.
[00:05:33.610]You want to start that in another set.
[00:05:36.360]So specifically for that backup path on that request.
[00:05:39.400]And once we do,
[00:05:40.233]we wanna allocate the slots for this backup path
[00:05:42.450]and a working path is in the first method.
[00:05:44.600]And if no paths are available again,
[00:05:46.580]you wanna increment the index by one
[00:05:48.267]and start the process all over again, where essentially,
[00:05:51.010]essentially we would like to output a working path
[00:05:53.340]and a backup path so we segment for that request.
[00:05:57.860]Now for the experiment itself,
[00:06:00.100]We use four main data sets
[00:06:02.740]where each data set a container of source
[00:06:05.430]and end destination
[00:06:07.710]node
[00:06:08.543]along with the bit rate demanding gigabits per second
[00:06:11.150]where each while also have,
[00:06:13.460]each request will also have their own single link failure
[00:06:16.320]to deal with as well.
[00:06:18.510]We also use figure one
[00:06:19.860]as our NSF network topology that we use
[00:06:22.390]and re-constructed a Hamiltonian cycle
[00:06:24.250]from figure one and to figure two for link based protection.
[00:06:29.570]So with our preliminary results here
[00:06:32.060]we like to relate the number of frequencies slots
[00:06:34.760]and fiber links used for each method
[00:06:36.920]in order to carry out the request.
[00:06:38.710]We use our proposed approach,
[00:06:40.280]which will be compared to path linked based algorithms
[00:06:42.500]to better understand our approach
[00:06:43.930]in terms of functionalities with the other methods.
[00:06:47.170]The frequency slots and fiber links
[00:06:48.665]that are used to carry out the requests
[00:06:50.540]during a single link failure
[00:06:52.540]was observed where our proposed approach
[00:06:55.220]was definitely comparable
[00:06:56.750]with both the link and path based protection algorithms.
[00:07:00.220]The slots used during a failure is very similar
[00:07:02.870]to the existing Lincoln path based algorithms.
[00:07:05.381]Thus we can see that even in figure, excuse me,
[00:07:08.970]in figure three and four and request two
[00:07:11.530]that the same number of slots and fiber links that are used
[00:07:15.430]are all are the same for all the methods
[00:07:17.220]to carry out the request properly
[00:07:19.150]during a single link failure.
[00:07:21.660]Therefore, our approach is very similar and comparable
[00:07:24.730]with the existing algorithms
[00:07:25.929]which provides dependable protection
[00:07:30.230]and our approach is reliable and committed
[00:07:32.110]to providing the protection needed
[00:07:34.080]during signaling failure through fiber links and slots used.
[00:07:37.680]In the future, we would have to apply each algorithm
[00:07:40.360]in a large elastic optical network based simulator
[00:07:43.600]for many different types of requests,
[00:07:46.510]while calculating the complexity spectrum utilization
[00:07:49.840]locking probability and resource utilization ratio
[00:07:53.500]of each algorithm for full optimal comparison.
[00:07:56.580]You will have to also take advantage
[00:07:58.260]of bandwidth variable transponders
[00:07:59.851]and considering the distance
[00:08:02.350]each path, each request takes and uses that information
[00:08:06.380]to determine the modulation formats
[00:08:08.640]to apply that to the working request.
[00:08:12.570]We also like to incorporate multiple network topologies
[00:08:15.760]for each method to provide again, a more concrete data.
[00:08:20.960]We would like,
[00:08:21.793]I would like to acknowledge professor Byrav Ramamurthy
[00:08:23.370]my UNL faculty mentor,
[00:08:25.730]and the UNL program scholars program,
[00:08:27.894]for all their, all of their guidance
[00:08:30.030]and help throughout this research project.
[00:08:33.620]Thank you very much for your time.
[00:08:34.860]And I would really appreciate you listening to me.
[00:08:36.317]Thank you.

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Comments

0 Comments

Investigation of Protection Failure Methods in WDM and Elastic Optical Networks

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments